Mimeme Attribute Classification using LDV Ensemble Multimodel Learning

One of the most common types social networking interaction is memes. Memes are innately multimodal, so studying and processing them a hot issue currently. This study's analysis DV dataset comprises classifying memes according to their irony, humour, motive, overarching mood. The effectiveness three different creative transformer-based strategies has been carefully examined. Dataset used here created by own meme data for this implementation hateful Out all our strategies, proposed ensemble model LDV obtained macro F1 score 0.737 humour classification, 0.775 motivation 0.69 sarcasm 0.756 overall sentiment meme.